Whereas psychophysicists may formulate hypotheses about appearance, they can only measure performance. Bias and imprecision in psychophysical data need not necessarily reflect bias and imprecision in perception. Sensory systems may exaggerate the differences between each item and its neighbors in an ensemble. Alternatively, sensory systems may homogenize the ensemble, thereby removing any apparent differences between neighboring items. Ensemble perception may be involuntary when observers attempt to report the identities of individual items. Conversely, when asked to make a (voluntary) decision about the ensemble as a whole, observers may find it very difficult to compute statistics that are based on more than a very small number of individual items. Modeling decisions about prothetic continua, such as size and contrast, can be tricky because sensory signals may be distorted before and/or after voluntarily computing ensemble statistics. With metathetic continua, such as spatial orientation, distortion is less problematic; physically vertical things necessarily appear close to vertical and physically horizontal things necessarily appear close to horizontal. Decision processes are corrupted by noise that, like distortion, may be added to sensory signals prior to and/or after voluntarily computing ensemble statistics.

中文翻译：

---

集合感知心理物理学中的五个二分法。

虽然心理物理学家可能会提出关于外表的假设，但他们只能衡量表现。心理物理数据的偏差和不精确不一定反映感知的偏差和不精确。感官系统可能会夸大整体中每个项目与其相邻项目之间的差异。或者，感官系统可以使整体均匀化，从而消除相邻项目之间的任何明显差异。当观察者试图报告单个项目的身份时，整体感知可能是非自愿的。相反，当被要求对整个集合做出（自愿）决定时，观察者可能会发现很难计算基于极少数单个项目的统计数据。模拟关于假体连续体的决策，例如大小和对比度，可能很棘手，因为感觉信号可能会在自愿计算整体统计之前和/或之后失真。使用复分解连续体，例如空间定向，失真问题较少；物理上垂直的东西必然看起来接近垂直，而物理上水平的东西必然看起来接近水平。决策过程会被噪声破坏，如失真，可能会在自愿计算整体统计之前和/或之后添加到感官信号中。